The logistics vehicle charging station location selection and routing problem with partial recharging and shared fleets

Growing global concern over environmental protection and climate change has led governments worldwide to promote electric vehicles to support carbon neutrality goals. However, limited charging infrastructure, particularly for urban logistics, remains a major barrier to the adoption of electric vehicles for delivery services. This study examines the logistics vehicle charging station location selection and routing problem with partial recharging and shared fleets. An electric vehicle charging model and a nonlinear energy consumption model are formulated to better represent real-world energy use and charging behavior. A bi-objective optimization model is proposed to minimize total operating costs and the required number of vehicles. To solve the model, a hybrid algorithm combining 3D k-means clustering and an improved multi-objective particle swarm optimization (IMOPSO) is developed. The 3D k-means clustering groups spatiotemporal customer data to support periodic resource allocation. The IMOPSO incorporates an enhanced update mechanism and elite selection to improve solution quality and convergence speed. In addition, a resource sharing strategy and a charging station insertion method are applied to further optimize vehicle deployment and station selection. The performance of IMOPSO is evaluated against the CPLEX solver, an improved non-dominated sorting genetic algorithm II, a flexible variable neighborhood search algorithm, and a multi-objective genetic algorithm with simulated annealing. A real-world case study in Chongqing City, China, assesses the proposed approach under sensitivity analysis of model parameters, five recharging levels, multiple resource sharing scenarios, and different collaboration modes. The results indicate that the proposed method supports efficient planning of urban delivery systems and charging infrastructure, contributing to a greener and more cost-effective logistics network.